From cc51a2f038098d1dda8bc5a761ba9869cffe5191 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Minh=20Nguy=E1=BB=85n?= <mxn@1ec5.org>
Date: Sun, 14 Jun 2015 10:18:04 -0700
Subject: [PATCH] Anchors in Wikipedia URLs

Corrected several issues with section anchors in Wikipedia URLs:

 - When normalizing a pasted-in URL into a tag value, try to decode the anchor.
 - Also when normalizing into a tag value, recognize variant paths such as zh.wikipedia.org/zh-hk/.
 - When the tag value contains an anchor, encode it before adding it to the link.
---
 js/id/ui/preset/wikipedia.js | 32 +++++++++++++++++++++++++-------
 1 file changed, 25 insertions(+), 7 deletions(-)

diff --git a/js/id/ui/preset/wikipedia.js b/js/id/ui/preset/wikipedia.js
index c215587a3..dbc0f5217 100644
--- a/js/id/ui/preset/wikipedia.js
+++ b/js/id/ui/preset/wikipedia.js
@@ -84,12 +84,22 @@ iD.ui.preset.wikipedia = function(field, context) {
 
     function change() {
         var value = title.value(),
-            m = value.match(/https?:\/\/([-a-z]+)\.wikipedia\.org\/wiki\/(.+)/),
-            l = m && _.find(iD.data.wikipedia, function(d) { return m[1] === d[2]; });
+            m = value.match(/https?:\/\/([-a-z]+)\.wikipedia\.org\/(?:wiki|\1-[-a-z]+)\/([^#]+)(?:#(.+))?/),
+            l = m && _.find(iD.data.wikipedia, function(d) { return m[1] === d[2]; }),
+            anchor;
 
         if (l) {
             // Normalize title http://www.mediawiki.org/wiki/API:Query#Title_normalization
-            value = m[2].replace(/_/g, ' ');
+            value = decodeURIComponent(m[2]).replace(/_/g, ' ');
+            if (m[3]) {
+                try {
+                    // Best-effort `anchordecode:` implementation
+                    anchor = decodeURIComponent(m[3].replace(/\.([0-9A-F]{2})/g, '%$1'));
+                } catch (e) {
+                    anchor = decodeURIComponent(m[3]);
+                }
+                value += "#" + anchor.replace(/_/g, ' ');
+            }
             value = value.slice(0, 1).toUpperCase() + value.slice(1);
             lang.value(l[1]);
             title.value(value);
@@ -102,14 +112,22 @@ iD.ui.preset.wikipedia = function(field, context) {
 
     i.tags = function(tags) {
         var value = tags[field.key] || '',
-            m = value.match(/([^:]+):(.+)/),
-            l = m && _.find(iD.data.wikipedia, function(d) { return m[1] === d[2]; });
+            m = value.match(/([^:]+):([^#]+)(?:#(.+))?/),
+            l = m && _.find(iD.data.wikipedia, function(d) { return m[1] === d[2]; }),
+            anchor = m && m[3];
 
         // value in correct format
         if (l) {
             lang.value(l[1]);
-            title.value(m[2]);
-            link.attr('href', 'http://' + m[1] + '.wikipedia.org/wiki/' + m[2]);
+            title.value(m[2] + (anchor ? ('#' + anchor) : ''));
+            try {
+                // Best-effort `anchorencode:` implementation
+                anchor = encodeURIComponent(anchor.replace(/ /g, '_')).replace(/%/g, '.');
+            } catch (e) {
+                anchor = anchor.replace(/ /g, '_');
+            }
+            link.attr('href', 'http://' + m[1] + '.wikipedia.org/wiki/' +
+                      m[2].replace(/ /g, '_') + (anchor ? ('#' + anchor) : ''));
 
         // unrecognized value format
         } else {